The Agent Skills Directory

[SAFE]: The skill is designed as a security auditing utility. All scripts and instructions focus on identifying and remediating common security misconfigurations in GitHub Actions workflows.
[EXTERNAL_DOWNLOADS]: The documentation references official GitHub security guides and well-known open-source security tools like actionlint and harden-runner from StepSecurity. These references are informative and point to trusted organizations.
[INDIRECT_PROMPT_INJECTION]: The auditing scripts (agent.py and process.py) ingest workflow files from the repository.
Ingestion points: Files matching .github/workflows/*.yml or .yaml (detected in scripts/agent.py and scripts/process.py).
Boundary markers: Not applicable, as the scripts perform static analysis rather than interpolating data into LLM prompts.
Capability inventory: The scripts are limited to file system read access for auditing and local file write access for report generation (json.dump). There is no network access or subprocess execution detected.
Sanitization: The scripts use yaml.safe_load() to parse workflow files, preventing YAML-based deserialization attacks.
[DATA_EXPOSURE]: The audit tools read local workflow configurations to generate a security report. This behavior is transparent, localized to the user's environment, and strictly aligned with the skill's stated purpose.

securing-github-actions-workflows